In [1]:
!pip install plotly==4.7.1 cufflinks
import plotly
print(plotly.__version__)
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
%matplotlib inline
import seaborn as sns
import plotly.express as px
import plotly.graph_objects as go
from sklearn.impute import SimpleImputer
# Basic packages
import pandas as pd, numpy as np, matplotlib.pyplot as plt, seaborn as sns, gc
from scipy import stats; from scipy.stats import zscore, norm, randint
import matplotlib.style as style; style.use('fivethirtyeight')
# Display settings
pd.options.display.max_rows = 400
pd.options.display.max_columns = 100
pd.options.display.float_format = "{:.2f}".format
random_state = 42
np.random.seed(random_state)
# Suppress warnings
import warnings; warnings.filterwarnings('ignore')
In [ ]:
print(plotly.__version__)
Malnutrition is a condition that results from eating a diet in which one or more nutrients are either not enough or are too much such that the diet causes health problems. It may involve calories, protein, carbohydrates, fat, vitamins or minerals. Not enough nutrients is called undernutrition or undernourishment while too much is called overnutrition.
People also often get infections and are frequently cold. The symptoms of micronutrient deficiencies depend on the micronutrient that is lacking. Source.
Import Data and Overview¶
In [2]:
data = pd.read_csv('malnutrition-estimates.csv')
data_by_country = pd.read_csv('country-wise-average.csv')
data.drop(['Unnamed: 0','ISO code','Survey Year','Source','Report Author','Notes','Short Source'], axis=1, inplace=True)
data.head()
Out[2]:
In [3]:
data.columns
Out[3]:
In [4]:
data.info()
In [5]:
data.describe().T
Out[5]:
Handling Missing values¶
In [6]:
# Check missing values in the dataframe
data.isnull().sum()
Out[6]:
In [7]:
columns = list(['Severe Wasting', 'Wasting','Overweight', 'Stunting', 'Underweight'])
print('Descriptive Stats before imputation for columns with missing values: \n', '--'*35)
display(data[columns].describe().T)
data['Wasting'].fillna(data['Wasting'].mean(), inplace=True)
data['Severe Wasting'].fillna(data['Severe Wasting'].mean(), inplace=True)
data['Overweight'].fillna(data['Overweight'].mean(), inplace=True)
data['Stunting'].fillna(data['Stunting'].mean(), inplace=True)
data['Underweight'].fillna(data['Underweight'].mean(), inplace=True)
print('Descriptive Stats after imputation: \n', '--'*35)
display(data[columns].describe().T)
Univariate Analysis¶
In [8]:
# Functions that will help us with EDA plot
def odp_plots(df, col):
f,(ax1, ax2, ax3) = plt.subplots(1, 3, figsize = (15, 7.2))
# Boxplot to check outliers
sns.boxplot(x = col, data = df, ax = ax1, orient = 'v', color = 'darkslategrey')
# Distribution plot with outliers
sns.distplot(df[col], ax = ax2, color = 'teal', fit = norm).set_title(f'{col} with outliers')
# Removing outliers, but in a new dataframe
upperbound, lowerbound = np.percentile(df[col], [1, 99])
y = pd.DataFrame(np.clip(df[col], upperbound, lowerbound))
# Distribution plot without outliers
sns.distplot(y[col], ax = ax3, color = 'tab:orange', fit = norm).set_title(f'{col} without outliers')
kwargs = {'fontsize':14, 'color':'black'}
ax1.set_title(col + ' Boxplot Analysis', **kwargs)
ax1.set_xlabel('Box', **kwargs)
ax1.set_ylabel(col + ' Values', **kwargs)
return plt.show()
Distribution plots¶
In [9]:
# Outlier, distribution for columns with outliers
boxplotcolumns = ['Severe Wasting', 'Wasting', 'Overweight', 'Stunting',
'Underweight']
for cols in boxplotcolumns:
Q3 = data[cols].quantile(0.75)
Q1 = data[cols].quantile(0.25)
IQR = Q3 - Q1
print(f'{cols.capitalize()} column', '--'*40)
count = len(data.loc[(data[cols] < (Q1 - 1.5 * IQR)) | (data[cols] > (Q3 + 1.5 * IQR))])
print(f'no of records with outliers values: {count}')
display(data.loc[(data[cols] < (Q1 - 1.5 * IQR)) | (data[cols] > (Q3 + 1.5 * IQR))].head())
print(f'EDA for {cols.capitalize()} column', '--'*40)
odp_plots(data, cols)
del cols, IQR, boxplotcolumns
Observation: There are outliers in the columns plotted above and we also plotted for values without outliers, hardly making any difference.
Multivariate Analysis¶
In [10]:
corr = data.corr()
mask = np.zeros_like(corr, dtype = np.bool)
mask[np.triu_indices_from(mask)] = True
sns.heatmap(corr, mask = mask, linewidths = .5 )#, annot = True)
Out[10]:
In [11]:
# Filter for correlation value greater than threshold
sort = corr.abs().unstack()
sort = sort.sort_values(kind = "quicksort", ascending = False)
display(sort[(sort > 0.7) & (sort < 1)])
Observation: As expected stunting, underweight and wasting are highly correlated.¶
Which country shows highest Underweight percentage? ---> Bangladesh¶
In [12]:
country = data.loc[:,['Country','Underweight']]
country['percunder'] = country.groupby('Country')['Underweight'].transform('max')
country = country.drop('Underweight',axis=1).drop_duplicates().sort_values('percunder', ascending=False).head()
fig = px.pie(country, names='Country', values='percunder', template='seaborn')
fig.update_traces(rotation=90, pull=[0.2,0.03,0.1,0.03,0.1], textinfo="percent+label", showlegend=False)
fig.show()
Which country shows highest Overweight percentage? ---> Albania¶
In [13]:
country = data.loc[:,['Country','Overweight']]
country['percunder'] = country.groupby('Country')['Overweight'].transform('max')
country = country.drop('Overweight',axis=1).drop_duplicates().sort_values('percunder', ascending=False).head()
fig = px.pie(country, names='Country', values='percunder', template='seaborn')
fig.update_traces(rotation=90, pull=[0.2,0.03,0.1,0.03,0.1], textinfo="percent+label", showlegend=False)
fig.show()
Which income class have highest underweight percentage? ---> Lower Middle Income¶
In [14]:
f,(ax1, ax2, ax3) = plt.subplots(1, 3, figsize = (15, 7.2))
sns.distplot(data['Underweight'], ax=ax1)
df_LM = data.loc[data['Income Classification'] == 'Lower Middle Income']
df_UM = data.loc[data['Income Classification'] == 'Upper Middle Income']
df_Low = data.loc[data['Income Classification'] == 'Low Income']
df_High = data.loc[data['Income Classification'] == 'High Income']
sns.distplot( df_LM['Underweight'],ax = ax2 , color = 'r')
sns.distplot( df_UM['Underweight'],ax = ax2, color = 'g')
sns.distplot( df_Low['Underweight'],ax = ax2, color = 'b')
sns.distplot( df_High['Underweight'],ax = ax2, color = 'y')
df = data.loc[:,['Income Classification','Underweight']]
df['maxunder'] = df.groupby('Income Classification')['Underweight'].transform('mean')
df = df.drop('Underweight', axis=1).drop_duplicates()
df = data.loc[:,['Income Classification','Underweight']]
df['maxunder'] = df.groupby('Income Classification')['Underweight'].transform('mean')
df = df.drop('Underweight', axis=1).drop_duplicates()
fig = sns.barplot(data=df, x='Income Classification', y='maxunder')
fig.set(xticklabels = ['LM', 'UM', 'Low', "High"])
plt.show()
In [15]:
df = data.loc[:,['Income Classification','Underweight']]
df['maxunder'] = df.groupby('Income Classification')['Underweight'].transform('max')
df = df.drop('Underweight', axis=1).drop_duplicates()
fig = px.pie(df, names='Income Classification', values='maxunder', template='seaborn')
fig.update_traces(rotation=90, pull=0.05, textinfo="percent+label", showlegend=False)
fig.show()
Underweight percentage in Least Developed Country vs Developed Country¶
In [16]:
f,(ax1, ax2, ax3) = plt.subplots(1, 3, figsize = (15, 7.2))
df_with_LDC = data.loc[data['LDC'] == 1]
df_with_DC = data.loc[data['LDC'] == 0]
sns.distplot(data['Underweight'], ax=ax1)
sns.distplot( df_with_LDC['Underweight'],ax = ax2 , color = 'r')
sns.distplot( df_with_DC['Underweight'],ax = ax2, color = 'g')
df = data.loc[:,['LIFD','Underweight']]
df['maxunder'] = df.groupby('LIFD')['Underweight'].transform('mean')
df = df.drop('Underweight', axis=1).drop_duplicates()
fig = sns.barplot(data=df, x='LIFD', y='maxunder', ax=ax3)
fig.set(xticklabels = ['Not LIFD', 'LIFD'])
plt.show()
Observation: Least developed country(green) has has more no of underweights and no in developed countries are way lesser considerably .
Low Income Food Deficient Countries Underweight percentages¶
In [17]:
f,(ax1, ax2, ax3) = plt.subplots(1, 3, figsize = (15, 7.2))
df_with_LIFD = data.loc[data['LIFD'] == 1]
df_with_NLIFD = data.loc[data['LIFD'] == 0]
sns.distplot(data['Underweight'], ax=ax1)
sns.distplot( df_with_LIFD['Underweight'],ax = ax2 , color = 'r')
sns.distplot( df_with_NLIFD['Underweight'],ax = ax2, color = 'g')
df = data.loc[:,['LIFD','Underweight']]
df['maxunder'] = df.groupby('LIFD')['Underweight'].transform('mean')
df = df.drop('Underweight', axis=1).drop_duplicates()
df = data.loc[:,['LIFD','Underweight']]
df['maxunder'] = df.groupby('LIFD')['Underweight'].transform('mean')
df = df.drop('Underweight', axis=1).drop_duplicates()
fig = sns.barplot(data=df, x='LIFD', y='maxunder')
fig.set(xticklabels = ['Not LIFD', 'LIFD'])
plt.show()
Underweight Analysis according to Income Group¶
In [18]:
data["Income Classification"].value_counts()
Out[18]:
Landlocked Developing Countries vs Small Island Developing states vs Others ---> Others¶
In [19]:
df = data.loc[:,['LLDC or SID2','Underweight']]
df['maxunder'] = df.groupby('LLDC or SID2')['Underweight'].transform('max')
df = df.drop('Underweight', axis=1).drop_duplicates()
fig = px.pie(df, names='LLDC or SID2', values='maxunder', template='seaborn')
fig.update_traces(rotation=90, pull=0.05, textinfo="percent+label", showlegend=False)
fig.show()
In [20]:
sns.kdeplot(data=data['Severe Wasting'], shade=True)
plt.title('Distribution of Sever Wasting percentages in countries')
plt.show()
We can notice from the distribution plot that the percentage values o as high as 12% in some countries.
From the scatter plot it can be noticed that high severe wasting percentages is mostly found for low survey sample sizes.
Severe Wasting - Wasting - Overweight - Underweight¶
In [21]:
sns.pairplot(data[['Severe Wasting','Overweight','Underweight', 'Stunting']])
plt.show()
In [22]:
sns.kdeplot(data=data['U5 Population (\'000s)'], shade=True)
plt.title('Distribution of U5 Population')
plt.show()
Relation of Underweight % to U5 population¶
In [23]:
fig = sns.scatterplot(data=data, x='Underweight', y='U5 Population (\'000s)')
fig.set(yticklabels=[])
plt.show()
It can be clearly noticed that for highest populations the underweight percentage is also quite high.
Which country shows highest underweight count? ---> India¶
In [24]:
df = data.loc[:,['Country','Underweight','U5 Population (\'000s)']]
df['underweight_count'] = (df['U5 Population (\'000s)'] * df['Underweight'])/100
df.drop(['Underweight','U5 Population (\'000s)'], axis=1, inplace=True)
df['undermean'] = df.groupby('Country')['underweight_count'].transform('mean')
df = df.drop('underweight_count', axis=1).drop_duplicates().sort_values('undermean', ascending=False).head()
fig = px.pie(df, names='Country', values='undermean', template='seaborn')
fig.update_traces(rotation=90, pull=[0.2,0.03,0.1,0.03,0.1], textinfo="percent+label", showlegend=False)
fig.show()
The reason behind this exceptional percentage may be due to the population count or any other factor as well.